Learn AI With Me — Week 2:
Sky Replacement & Lighting Correction (Ethically)

Learn AI With Me — Week 2: Sky Replacement & Lighting Correction (Ethically)
Last week, I started this series with basic photo cleanup using AI.
This week, we’re moving into something a little more advanced — but still practical, still ethical, and still grounded in real‑world commercial real estate: sky replacement and lighting correction.
This is one of the most common AI photo enhancements, and also one of the easiest places to accidentally cross the line into misrepresentation.
In Week 2, the goal wasn’t to make the photo “pretty.”
The goal was to make it
accurate, realistic, and slightly better — without changing the property itself.
Here’s how it went.
1. The Original Photo
I started with a simple exterior shot taken on an overcast day.
Flat lighting, dull sky — exactly the kind of photo we all end up with when we’re out in the field and the weather isn’t cooperating.
2. The First Prompt (Conservative & Ethical)
I began with a very safe, very conservative prompt in Nano‑Banana:
“Replace the overcast sky with a natural, lightly clouded sky that matches the soft lighting in the scene. Improve brightness, contrast, and color accuracy slightly. Do not alter the building, signage, people, shadows, or any structural elements. Keep the property exactly as it appears in real life.”
This prompt is intentionally cautious.
It tells the AI:
- don’t change the building
- don’t change the dimensions
- don’t change the features
- don’t change the condition
- don’t add or remove anything
- keep the lighting realistic
This is the foundation of ethical AI photo editing in CRE.
3. The First Enhanced Photo (Technically Correct… But Subtle)
Nano‑Banana followed the instructions perfectly — maybe too perfectly.
(Insert first enhanced photo here)

The result was technically correct, but honestly, it didn’t look that different.The sky was slightly improved, the lighting was slightly better, but the overall effect was minimal.
And that’s normal.
AI follows conservative prompts literally. If you tell it to be subtle, it will be subtle.
4. Adjusting the Prompt (Still Ethical, But With More Punch)
Since the first version didn’t move the needle enough, I refined the prompt:
“Replace the overcast sky with a natural, lightly clouded sky that adds a bit more brightness but still matches the soft lighting. Increase overall brightness and contrast moderately. Improve color accuracy and clarity. Do not alter the building, signage, people, shadows, or any structural elements.”
This version gives the AI permission to:
- brighten the sky
- increase contrast
- improve clarity
- add a bit more visual pop
…but still without touching the building.
This is the key rule in commercial real estate: You can change the weather, but you cannot change the property.
5. The Final Enhanced Photo (Realistic & Accurate)
The second result was much better:
(Insert final enhanced photo here)

- brighter sky
- better lighting
- improved contrast
- more accurate color
- still realistic
- still true to the property
This is exactly what ethical AI enhancement should look like.
6. What You Can Ask AI to Do (and What You Shouldn’t)
AI can do a lot more than subtle sky replacement.
If you want, you can ask it to:
- make it sunny
- make it snow
- make it rain
- make it nighttime
- add clouds
- remove clouds
- change the season
And all of that is fine as long as the building stays exactly the same.
The structure, dimensions, features, and condition must remain untouched. That’s the line.
7. Tools You Can Use
For this exercise, I used:
- Nano‑Banana (ChatGPT 5)
But you can also use:
- Use.AI
- Adobe Express
- Luminar Neo
- Photoleap
The tool matters less than the prompt and the ethics.
8. What I Learned This Week
A few takeaways from Week 2:
- AI editing is iterative — the first result is rarely the final one
- Conservative prompts produce conservative results
- You can dial up the effect while staying ethical
- The building must never change
- Sky replacement is powerful, but subtlety is your friend
- You can make dramatic weather changes, but accuracy matters
- AI is helping me learn techniques I wouldn’t have tried on my own
This is exactly why I’m doing this series — to learn, to experiment, and to document the process honestly.
Next Week: Object Removal (Ethically)
Week 3 will focus on removing distractions:
- trash cans
- cones
- cars
- clutter
- temporary objects
And again — the rule stays the same: Remove distractions, not features.
See you next week.












